Intelligent driving scenario adaptive teaching method and system based on reinforcement learning

By adopting intelligent driving scenario adaptive teaching methods based on reinforcement learning in the intelligent driving system, using two-way long and short-term memory networks and multi-head self-attention mechanisms for scene recognition and decision-making model training, the problems of inaccurate scene recognition and poor generalization capabilities in the existing technology are solved, and efficient and personalized driver training and system performance improvement are achieved.

CN119417671BActive Publication Date: 2025-04-01YIXIAN INTELLIGENCE
2 Cites 1 Cited by

Patent Information

Application Number
CN202510018846.6
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2025-01-07
Publication Date
2025-04-01
Estimated Expiration
2045-01-07

AI Technical Summary

Technical Problem

The existing adaptive teaching methods for intelligent driving scenarios based on reinforcement learning have problems such as inaccurate scene recognition, poor model generalization ability, and single feedback on driving guidance information, which is difficult to effectively improve the driver's learning efficiency.

Method used

Adaptive teaching method for intelligent driving scenarios based on reinforcement learning is adopted, and multiple scenario recognition models are trained by collecting intelligent driving historical data, and combining two-way long and short-term memory networks and multi-head self-attention mechanisms to conduct scene classification and intelligent driving decision model training. The driving data is collected in real time, the current scene type is determined through the scene recognition model, the corresponding decision model is selected, the standard driving behavior sequence is generated, the driving behavior deviation value is calculated, the teaching feedback strategy is determined, and the decision model parameters are updated.

Benefits of technology

It improves the efficiency and pertinence of intelligent driving training, improves drivers' driving skills, realizes adaptive learning and continuous optimization, and improves the performance and driving experience of intelligent driving systems.

✦ Generated by Eureka AI based on patent content.
Patent Text Reader

Abstract

The present invention provides an intelligent driving scene adaptive teaching method and system based on reinforcement learning, which relates to the field of reinforcement learning technology, including training multiple scene recognition models based on historical data, and classifying the historical data into scenes to generate a labeled training sample set. The intelligent driving decision model for different scene types is trained using a reinforcement learning algorithm. In the actual driving process, the current driving scene is identified based on the real-time collected vehicle position, speed, acceleration and other information, and the corresponding decision model is selected to output a standard driving behavior sequence. The actual behavior of the driver is compared with the standard behavior, the deviation value is calculated, and the teaching feedback strategy is determined based on the deviation value, the driving guidance information is output in real time, and the driver feedback is used to update the decision model. The present invention can provide personalized teaching according to different driving scenarios, improve the driver's driving skills, and enhance driving safety.
Need to check novelty before this filing date? Find Prior Art